metadata
license: creativeml-openrail-m
widget:
- text: amazing
- text: a photo of
- text: a sci-fi
- text: a portrait of
- text: a person standing
- text: a boy watching
Under Development
Major improvements from v1 will include:
- 25% more variations
- more capable of generating story-like prompts, using contrastive search instead of greedy search
- cleaned training data
- removed prompts that generate images with nsfw scores >= 0.5
- removed duplicates (including prompts that differ by capitalization and punctuations)
- removed punctuations in random places